02:42
<heycam>
Hixie, typo in most recent commit (you have "it's" instead of "its")
02:44
<Hixie>
where is it?
02:44
<Hixie>
i don't recall what the last commit was :-)
02:45
<heycam>
in the Obsolete features / Conformance checkers section
02:48
<heycam>
http://lists.w3.org/Archives/Public/public-html-diffs/2008Dec/0121.html
02:49
heycam
afk for a bit
02:56
<Philip`>
"The presence of a profile attribute on the head element, if it's value is ..."
03:01
<Hixie>
i fixed it as soon as heycam said "in the Obsolete features / Conformance checkers section" :-)
09:23
<Hixie>
what does one call a statement that doesn't modify something? as opposed to a statement that _does_ modify something
09:25
<heycam>
idempotent?
09:26
<heycam>
referentially transparent?
09:26
<Hixie>
idempotent statements can still change things, they are just guaranteed to not cause further changes if executed again
09:26
<Hixie>
referentially transparent might be it...
09:26
<heycam>
ah yes
09:27
<heycam>
side-effectless
09:27
heycam
cooks
09:28
<Hixie>
referentially transparent seems right
09:28
<Hixie>
although...
09:29
<Hixie>
UPDATE TABLE x SET a=0 WHERE 0==1; is referentially transparent
09:29
<Hixie>
and i don't want to include that
09:33
<roc>
"pure" is sometimes used
09:34
<roc>
"pure function" maybe
09:34
<roc>
a statement that changes no state and does not return a value (i.e., is not an expression) could also be referred to as a "no-op"
09:38
<Hixie>
i used "referentially transparent" but refer only to "the statement's main verb" (this is about sql)
09:38
<Hixie>
and then i give an example to show what i mean by "main verb"
09:39
<roc>
"has no side-effects" is probably more meaningful to more people
09:40
<Hixie>
yeah i considered that but i was worried people might think that "DROP TABLE foo" had no side-effects, only one primary effect.
09:41
<roc>
seems much more likely that people will have no clue what "referentially transparent" means
09:42
<Hixie>
well that's why i include the example :-)
09:43
<Hixie>
here, what do you think of this: http://www.whatwg.org/specs/web-apps/current-work/#modifications-fail-if-read-only
09:43
<Hixie>
step 6.2
09:49
<roc>
hmm
09:49
<roc>
actually "referential transparency" is not what you want here
09:49
<roc>
since, strictly speaking, functions that read mutable global state are not referentially transparent
09:49
<Hixie>
good point
09:50
<Hixie>
what's the right term?
09:50
<Hixie>
what i want is the accessor version of "immutable"
09:50
<Hixie>
i can think of half a dozen ways to describe a data structure that can't be modified, but no way to describe a function that doesn't modify a data structure
09:50
<Hixie>
go figure
09:51
<roc>
maybe "can modify the database" is as good as it gets
09:52
Hixie
removes the parenthetical
09:52
<Hixie>
i'm sure there's a word
09:53
<Hixie>
but "can't modify" will have to do for now
10:22
<MikeSmith>
wasn't mpt particularly interested in notifications?
10:32
<Hixie>
MikeSmith: the gears team were the only people who said anything about implementing it as far as i know
10:32
<Hixie>
MikeSmith: and they want full-on HTML notifications
10:32
<MikeSmith>
OK
10:33
<Hixie>
MikeSmith: which i'm not willing to spec if they're the only implementation, because that's heavy duty stuff
10:33
<MikeSmith>
yeah, I can understand that
10:34
<MikeSmith>
Hixie: If there is significant interest from them in having taking a spec for it to the Webapps WG, I would kinda hope to get word on it sooner rather than later
10:35
<MikeSmith>
Is there somebody in particular you can suggest I ping about it?
10:36
<MikeSmith>
or course it'd be great to have somebody willing to commit to being the editor for it
10:36
<Hixie>
i don't know of any interest from that team to write a spec yet
10:36
<Hixie>
but i'm trying to encourage them to work with the w3c
10:37
<Hixie>
the thing is there's no point having a spec if there's only one company interested in it
10:37
<Hixie>
because once the other vendors start getting interested, they'll want to change it
10:37
<MikeSmith>
yep
10:40
<MikeSmith>
I think Andrei's work on the Geolocation draft has proven that when there is interest among multiple vendors, work on standard can get done relatively quickly and successfully
10:40
<MikeSmith>
with multiple implementations
10:49
<yecril71>
Modifying the fingerprint on the advertising site would only disrupt the site.
10:50
<yecril71>
It would not cause the downloaders to download the trojan � since the original file has not been modified.
10:51
<Hixie>
yecril71: just change the link to point to the trojan on some other site. or upload the trojan straight to that site. or do both.
10:52
<yecril71>
That would be easy to recognise since the browser starts downloading from another domain.
10:52
<roc>
the problem with HTML notifications is that that makes it impossible to integrate with platform notification frameworks such as Growl or libnotification
10:53
<Hixie>
roc: yeah i'm not really a big fan of it. i've been trying to discourage them (unsuccessfully)
10:56
<yecril71>
(it seems the fingerprint could cover the URL as well)
10:56
<Hixie>
?
10:56
<yecril71>
Oops, that is irrelevant, sorry.
11:45
<Philip`>
Hixie: Judging by http://lists.whatwg.org/pipermail/commit-watchers-whatwg.org/2008/001779.html , the .png file probably needs its svn:mime-type property set to something non-textual
11:46
<Hixie>
how do i change that?
11:49
<Philip`>
svn propset svn:mime-type image/png content-venn.png
11:49
<Philip`>
... I think
11:50
<Philip`>
(Apparently anything not starting with "text/" is considered to be binary, and prevents textual merging and newline conversion and all that stuff)
11:54
<Hixie>
done
11:55
<annevk>
data:text/xml,<test xmlns="&gt;">test</test> should this work?
11:55
<annevk>
"The attribute's normalized value MUST be either a URI reference — the namespace name identifying the namespace — or an empty string."
12:01
<annevk>
http://www.w3.org/TR/xforms11/#fn-digest should we have something like this for password fields?
12:02
<Hixie>
what problem does it solve?
12:03
<annevk>
sending passwords more securely over the wire without having to resort to scripting aiui
12:04
<Hixie>
what does that protect against?
12:05
<annevk>
knowledge of the original string used which may be in use on other locations
12:05
<Hixie>
ah
12:05
<Hixie>
seems like SSL is a better solution to that
12:06
<annevk>
blog software etc. is not going to switch to SSL, is it?
12:06
<Hixie>
(also, if you can sniff the hash for one site, seems likely that you can sniff the hash for another site)
12:06
<Hixie>
if you're using the same password for a blog as for something which matters, you're pretty screwed anyway
12:07
<Hixie>
the other problem is that it would mean one of two things
12:07
<Hixie>
either the site has to store the actual password instead of a hash
12:08
<Hixie>
or the site has to support both legacy login and new login
12:08
<Hixie>
the former increases the risk of the site being hacked and the passwords leaked, which is a bigger risk than sniffing in practice
12:08
<Hixie>
the latter seems like an unlikely thing for sites to do
12:22
<yecril71>
If a number property cannot be coerced to an IDL number because it is too big,
12:22
<yecril71>
the property accessor returns 0.
12:22
<yecril71>
But it looks rather like NOT_SUPPORTED_ERR than 0 to me.
12:23
<yecril71>
Because the property contains a valid number, only its value is inaccessible to the script.
12:28
<heycam>
yecril71, do you have some examples/tests that show this?
12:30
<yecril71>
It is specified in 2.8.
12:30
<yecril71>
I do not know how implementations behave.
12:31
<heycam>
2.8?
12:31
<heycam>
oh of html5?
12:33
<heycam>
hmm those algorithms in 2.4.3 Numbers don't say what to do with overflow
12:34
<heycam>
or rather, the section on reflecting dom attributes don't say that such overflows are not "successful"
12:35
heycam
→ bed
12:41
<yecril71>
Numbers do not overflow in HTML5 at all.
12:42
Hixie
wonders how to define what a valid floating point number represents without resorting to the parsing algorithm
12:48
<yecril71>
A valid floating point number represents an abstract object with the following properties:
12:49
<annevk>
Hixie, ABNF!
12:49
<yecril71>
sign, value (as integer), exponent (as integer).
12:49
<Hixie>
annevk: that doesn't define what it means, only what the syntax is
12:49
<Hixie>
yecril71: that doesn't say what a particular string represents
12:49
gsnedders
notes RFC 3339 doesn't define what the string means
12:50
<yecril71>
Figuring out what a particular string represents is called parsing.
12:51
<yecril71>
Do you want to parse without resorting to the parsing algorithm?
12:51
<Hixie>
no, i want to define what a string means for authors, without them having to understand a fifteen step parsing algorithm
12:52
Hixie
does so and checks in the result
12:55
<yecril71>
Authors are expected to understand what a decimal expansion is.
12:55
<yecril71>
The exponent is a notation for multiplication.
12:55
<Hixie>
nothing in the spec actually said that until abotu 2 minutes ago
12:56
Hixie
wonders how to resolve zcorpan's feedback regarding the parsing of numbers
12:56
<Hixie>
IE's parsing algorithm is wacky
12:56
gsnedders
places bet it's saner than its HTTP parsing
12:58
<Hixie>
IE checks if the next character is [a-zA-Z] or [\192-\214] or
12:58
<Hixie>
> [\216-\246] or [\248-\501] or [\506-\535] or [\592-\680] or \902 or
12:58
<Hixie>
> [\905-\906] or \908 or [\910-\929] .... etc, and if so, drop the
12:58
<Hixie>
> attribute.
12:58
Hixie
doesn't understand what determines membership in that set
12:58
<takkaria>
gsnedders: IE's number parsing algorithm parses HTTP? :P
12:58
<gsnedders>
takkaria: :P
12:58
<gsnedders>
takkaria: I haven't even looked a number parsing in HTTP
12:59
<gsnedders>
But that description so far seems saner than parts of the HTTP behaviour
13:01
<yecril71>
I would say IE treats letters as invalid but allows symbols at the end.
13:02
<Hixie>
define "letters"?
13:02
<yecril71>
You already have.
13:02
<Hixie>
as what?
13:03
<yecril71>
[a-zA-Z] or [\192-\214] or�
13:03
<Hixie>
or what? the list above is incomplete.
13:03
<Hixie>
and i certainly don't want to list the entire unicode table with "yes" or "no" for each character.
13:04
<yecril71>
You can just refer to Unicode.
13:05
<Hixie>
...and say what?
13:06
<yecril71>
Letters are invalid, symbols are ignored.
13:06
<Hixie>
what do you mean by "letters"?
13:07
<yecril71>
Characters used to spell words.
13:07
<Hixie>
is U+8273 a letter?
13:07
<Hixie>
or U+2874
13:07
<Hixie>
or any other character?
13:07
<Hixie>
we need to define this precisely
13:07
<Hixie>
in a way that matches IE
13:07
<Hixie>
if we are to do this
13:08
<Hixie>
just saying "letters" is inadequate
13:08
<yecril71>
U+8273 is a letter.
13:09
<annevk>
Is "Letter" a character class in Unicode?
13:09
<yecril71>
I guess so.
13:09
<annevk>
e.g. http://msdn.microsoft.com/en-us/library/20bw873z.aspx suggests Microsoft might use something like that
13:10
<Hixie>
guessing isn't really going to cut it here
13:10
<yecril71>
The meaning of punctures depends on the language.
13:11
<yecril71>
I think they should not be treated as letters.
13:11
<Hixie>
what we think isn't really relevant here
13:11
<Hixie>
if we're trying to match IE
13:11
<Hixie>
annevk: yeah that might be what they're using. i'd need to check the ranges.
13:12
<yecril71>
Note that IE supports BMP characters only.
13:16
<annevk>
http://www.unicode.org/Public/UNIDATA/PropList.txt code points marked L& are that class prolly
13:16
annevk
is not sure whether the list is complete, though it looks like it
13:19
<yecril71>
The list is expected to grow as new characters are identified.
13:19
annevk
whois test
13:20
<annevk>
wtf, HTTP stopped working
13:20
<annevk>
yecril71, sure
13:23
<Philip`>
Hixie: Should "-0" be a valid signed integer?
13:24
<Philip`>
Hixie: If it is, it doesn't represent the negative number represented in base ten by the string "0", it just represents zero
13:24
<Hixie>
hm
13:27
<yecril71>
How can a script detect HTTP 404 in a window it is driving?
13:28
<Hixie>
Philip`: better?
13:28
<yecril71>
responseXML is called XMLDocument in mshtml
13:28
<Hixie>
yecril71: the server knows it sent a 404, so it can put a marker in the page
13:29
<yecril71>
But the marker can be faked.
13:30
<Philip`>
Hixie: Sounds okay now
13:30
<Hixie>
if the server is faking the marker, it can fake the 404...
13:30
<Philip`>
(where "okay" means "probably about correct", as opposed to "understandable and helpful to normal people")
13:30
<yecril71>
The server cannot fake the 404.
13:31
<Hixie>
The server can fake the 404.
13:31
<yecril71>
If the server returns 404, it is genuine.
13:31
<jwalden>
<?php header("HTTP/1.1 404 Not Found"); ?> or somesuch
13:31
<Philip`>
Hixie: Valid floating point numbers don't really represent m*10^e - they represent a floating point approximation of that number
13:31
<Hixie>
heycam: wow, congratulations on your appointement in the svgwg
13:32
<Hixie>
Philip`: how do you mean?
13:32
<jwalden>
Hixie: 2, not 10
13:32
<yecril71>
jwalden: that is still genuine.
13:32
Philip`
wonders why he is able to send messages to this channel but has not received any for the past few minutes
13:32
<jwalden>
not sure I understand "genuine"
13:33
Philip`
thus relies on krijnh's logs to allow bidirectional communication
13:33
<Hixie>
yecril71: if your point is that the server can make mistakes, then yes, but that's a bug in the server. it can just as easily make the mistake of sending back 404 for a page that it shouldn't send back a 404 for.
13:33
<Hixie>
jwalden: 5e3 is 5 x 10 ^ 3
13:33
<Hixie>
jwalden: not 5 x 2 ^ 3
13:34
<yecril71>
The server has an ultimate authority on the resources it serves.
13:34
<jwalden>
maybe I don't have enough context; I was assuming this was in reference to idealized floating-point numbers
13:34
<yecril71>
If it says it is not there, it is not there.
13:34
<yecril71>
It is like immaculate conception.
13:35
<yecril71>
God cannot lie, and the server cannot fake 404.
13:35
<yecril71>
Once it returns 404, the resource is REALLY NOT THERE.
13:35
<Hixie>
god is omnipotent, of course he can lie :-)
13:36
<yecril71>
God cannot lie in the sense that whatever he says, becomes truth at the very moment.
13:37
<Hixie>
so what happens if god says "This sentence is a lie"?
13:37
<yecril71>
Our universe transforms to an alternative one.
13:38
<Hixie>
(or what happens if you are trying to find a picture of god, and you go to http://example.org/god, and the server returns a picture of god, with the status code 404?)
13:38
<takkaria>
"please leave your sense of logic at the door" has never been more applicable
13:39
<yecril71>
The client script cannot detect it is 404.
13:39
<Philip`>
Hixie: I mean that e.g. "1e-1" doesn't represent 1*10^-1, it represents (something)*2^(something) which is a different number
13:40
<Hixie>
Philip`: no, it represents 0.1. Now the browser might have a hardware limitation that prevents it from accurately representing that number efficiently, and thus it might not in fact treat it as 0.1 but as some rough approximation of 0.1, but the string itself actually represents 0.1.
13:42
<Philip`>
I guess that's just a matter of perspective :-)
13:43
<Hixie>
Philip`: put it this way. if a browser implemented html5 using bcd representations, would it be non-conforming?
13:45
<Philip`>
Hixie: Probably, because ECMAScript requires IEEE 754 double semantics
13:46
<Hixie>
certainly when exposing the numbers to ECMAScript they'd need to be converted to base-2 representations
13:46
<Hixie>
but we're not talking about JS APIs here
13:46
<Hixie>
anyway
13:46
<Hixie>
bed time
13:46
<Hixie>
nn
13:46
<takkaria>
I thought BCD was something they made up to waste time in computing classes
13:47
<yecril71>
BCD is for calculating your tax returns.
13:47
<Philip`>
Some CPUs have special instructions for BCD computations
13:47
<gsnedders>
What's BCD?
13:47
<yecril71>
And the balance of your bank account.
13:47
<yecril71>
Binary Code for Digits.
13:47
<Philip`>
gsnedders: Binary Coded Decimal
13:48
<takkaria>
fixed-point, surely, not BCD?
13:48
<Philip`>
gsnedders: which is where you take a decimal number, then represent each digit as 4 binary digits
13:48
<gsnedders>
fun
13:48
<Philip`>
gsnedders: (so e.g. a 32-bit register could represent a 8-digit decimal number)
13:48
<gsnedders>
Yeah
13:49
<Philip`>
It makes arithmetic more fun
13:49
<yecril71>
IMB invented BCD for financial calculations on mainframes.
13:49
<yecril71>
s/IMB/IBM/
13:50
<yecril71>
in order to minimize roundoff errors.
13:51
<yecril71>
There is no way the page can know if the document has been printed or not.
13:51
<yecril71>
Short of revieweing a scan of the printout, that is.
13:52
<Philip`>
takkaria: You'd want fixed-point rather than floating-point, but I guess you'd want it to be decimal fixed-point rather than binary fixed-point so that you can convert to/from human-readable decimal representations with much less cost
13:53
gsnedders
needs to make notes on 100 pages per day to get through all the texts for his English dissertation before term starts again
13:53
<Philip`>
takkaria: (because converting a binary number to a decimal string is probably very expensive when the numbers have a dozen digits and your CPU is made of cheese and string)
13:54
<takkaria>
true
13:55
Philip`
once made a three-digit binary-to-decimal converter at school using simple NAND/NOR/etc chips, and it took two whole breadboards
13:59
<Philip`>
(I probably should have just used a BCD chip, but that would have been no fun)
13:59
<gsnedders>
Your school had BCD chips?
14:03
<gsnedders>
MikeSmith: Music suggestions?
14:03
<yecril71>
Footnotes are different from asides.
14:04
<yecril71>
Asides provide illustrative content,
14:04
<yecril71>
foot notes provide technical content that is no fun to read.
14:04
<MikeSmith>
gsnedders: MF DOOM, MadLib, .. MadVillian
14:05
<MikeSmith>
Yesterday's New Quintet
14:06
<MikeSmith>
Clutchy Hopkins
14:08
<Philip`>
gsnedders: Yes - 7441 BCD-to-decimal decoder, I think
14:10
<Philip`>
yecril71: The Hitchhiker's Guide disproves your claim that footnotes are no fun to read
14:10
<jwalden>
BCD finds use in TI-8[34].* calculators, among others, unless the machines have changed since the 83+ times
14:10
gsnedders
marks MikeSmith suggestions for later and tunes into gothic metal on last.fm
14:11
<MikeSmith>
Flann O'Brien proved some things about footnotes a while before that
14:12
<Philip`>
MikeSmith: But I've never read anything he wrote, whereas I have read HHGTTG, so I can only use the latter as evidence :-p
14:13
<takkaria>
try reading John Stuart Mill's Utilitarianism for an amusing footnote, and that was the 1800s
14:13
gsnedders
still hasn't worked out whether we're markup geeks or Eng.lit. geeks
14:13
<takkaria>
I'm a philosophy geek ^_^
14:14
<gsnedders>
takkaria: You're crazy.
14:14
<takkaria>
no, just bored :(
14:14
<yecril71>
What is fun in "op.cit., 100-157?"
14:15
<MikeSmith>
yecril71: stop, stop, you're making me do bellylaughs
14:16
<yecril71>
I shall consider stopping as soon as I get a decent footnote element.
14:17
<MikeSmith>
Philip`: you have something to look forward to if you've never read Flann O'Brien
14:18
<Philip`>
MikeSmith: That's only something to look forward to if I ever plan to read him in the future
14:18
gsnedders
goes back to writing notes on Nabokov
14:18
<MikeSmith>
yecril71: I recommend you stick to telling jokes like "op.cit., 100-157?".. that one's a classic. You should be a professional footnote comedian.
14:19
<Philip`>
That seems a very specialised profession
14:19
<Philip`>
LaTeX makes it a bit tricky to have footnotes inside footnotes, which always annoys me
14:20
<MikeSmith>
gsnedders: I've given up on music and now I just listen to old episodes of CBS Radio Mystery Theater featuring E.G. Marshall.. "the theater of the imagination"
14:20
<gsnedders>
"He felt both sorry and repelled but, realizing that the material, apart from its one specific function, had no potentional whatever, he kept doggedly at his chor"
14:20
<gsnedders>
*chore
14:21
<MikeSmith>
Philip`: if you plan to try psilocybin or LSD in the future, I recommend that you consider trying it in combination with reading At-Swim-Two-Birds
14:22
gsnedders
tries to imagine Philip` trying such things
14:22
gsnedders
fails miserably
14:26
<yecril71>
It is not possible to getElementById an element in a disconnected subtree,
14:27
<yecril71>
therefore the uniqueness restriction of identifiers in disconnected subtrees
14:27
<yecril71>
arguably serves no purpose.
14:27
<Philip`>
MikeSmith: I think the closest I'll ever get to that is trying BSD
14:31
<annevk>
yecril71, fwiw, IRC is not the best place to comment on the spec (I might have mentioned this already...)
14:32
<annevk>
yecril71, that is, unless Hixie is around and happens to be editing that part of the specification, it's generally a waste of time
14:34
<MikeSmith>
Philip`: I guarantee you the Berkeley guys who came up with BSD did their fair share of ther *SD and psilocybin. Billy Joy needed the holy ghost to be able to write his TCP/IP implementation
14:35
<MikeSmith>
yecril71: I think your IRC comments need more footnotes, preferably humorous ones.
14:38
<Philip`>
MikeSmith: That's not possible unless IRC has explicit syntax to represent footnote semantics
14:40
<yecril71>
Hixie has encouraged me to share my comments here.
14:40
<yecril71>
This way, he wastes my time by sending irrelevant information to me,
14:41
<jwalden>
probably to stimulate discussion before a proposal, I suspect
14:41
<yecril71>
while I do not waste his.
14:41
<jwalden>
email discussion is no fun for hashing out basics, too slow
14:41
<MikeSmith>
Philip`: I'm already on top of it. I'm on the #irc channel the master IRC network alerting J. Oikarinen about his lack of foresight
14:44
<Dashiva>
MikeSmith: Are you writing a XML-based replacement as well?
14:44
<Dashiva>
Atom is a good starting point
14:44
<takkaria>
the IRC protocol is a wonderful mess that deviates significantly from its documentation
14:45
<annevk>
Dashiva :D
14:45
<yecril71>
I imagine Hixie can scan the logs if he wishes to, provided the log machine works.
14:46
<yecril71>
So it is not quite like typing into the void.
14:46
<gsnedders>
takkaria: I think that's an understatement
14:47
<annevk>
oh, he has his own logs, but he rarely replies to random banter :)
14:48
<yecril71>
It is me who is replying here.
14:49
annevk
shrugs
14:52
<MikeSmith>
Dashiva: I'm writing it in Lisp
14:53
<takkaria>
s-expressions for IRC actually aren't a bad idea
14:54
<yecril71>
Playing jingles in the backround of spoken text does not seem like a good idea.
14:54
<yecril71>
It can make the text significantly harder to understand.
14:55
<gsnedders>
Why are there so many good Finnish bands?
14:56
<yecril71>
Because the Finns have long nigths with nothing to do :-)
14:56
<yecril71>
nights
15:04
<yecril71>
CSS behaviors, as implemented by Internet Explorer, are quite convenient.
15:05
<yecril71>
On the other hand, they require scripts to run anyway,
15:05
<yecril71>
and once scripts can run, the events can be attached using a setup script.
15:06
<yecril71>
This would require the CSS selector API.
16:02
<Philip`>
Dashiva: XMPP already supports multi-user chat, so that'd probably be a better starting point
16:02
<Philip`>
or maybe it's already the finishing point
16:02
<Philip`>
but anyway it uses XML so it must be good
16:03
<Philip`>
(Never mind that you can basically destroy an entire channel by sending namespace-ill-formed XML that the server will blindly retransmit to every other user in the channel and cause their connections to drop)
16:13
<jwalden>
that's a server bug, no?
16:13
<jwalden>
or so I'd expect
16:16
<Philip`>
Yes (in ejabberd) - it's fixable, but not trivially, and it's the kind of bug that XML seems to encourage
16:17
<Philip`>
(It's non-trivial because you need a namespace-aware self-contained-XML-fragment serialiser, and you need it written in Erlang)
16:20
<annevk>
hmm, class names are case-insensitive?
16:20
<annevk>
oh, they're not
16:36
<hdh0>
in 4.5.13.1, "because the order is not particular important" should be "particularly"
21:09
zcorpan_
ponders about the interactiveness of <details>
21:10
<zcorpan_>
Hixie: shouldn't <legend> ban interactive descendants if the parent is <details>?
22:03
<heycam>
thanks Hixie (re svgwg)
22:04
<heycam>
Philip`, i'd say that "-0" shouldn't be a signed integer, one reason for which would be that the integral types in Web IDL don't include a "-0" value
22:04
<heycam>
(though float does)
22:06
<Philip`>
heycam: There already isn't a one-to-one mapping between strings and values, e.g. "01" and "1" both mean the same number, so is it bad if "-0" and "0" both mean the same number too?
22:11
<heycam>
i don't think having multiple strings mapping to the same number is a problem. but the fact that "01" means the same regardless of whether the type is a long or a float, but "-0" means different things, might be trouble.
22:11
<heycam>
well not trouble, ...
22:11
<heycam>
but unexpected
22:12
<heycam>
(though i expect the number of authors to know about negative zeroes in floats to be small anyway)
22:18
<Philip`>
heycam: As far as I can tell, "-0" means the same as "0" in HTML5's definition of floating point numbers, because it's defined in terms of ideal real numbers
22:18
<Philip`>
(and ideal real numbers don't have a -0)
22:20
<heycam>
oh
22:20
<heycam>
so html5 parsing of floats is different from ecma-262 parsing of floats
22:22
<Philip`>
HTML5 parsing of floats doesn't even allow "1e+1"
22:22
<Philip`>
(It only allows "1e1" and "1e-1")
22:24
<Philip`>
which is probably going to be horribly confusing for anyone who's generating HTML from a language that serialises large floating point numbers with "...e+...", i.e. pretty much any language
22:24
<heycam>
yeah, i think it would be better if they parsed the same
22:24
<heycam>
is there a particular reason to disallow "1e+1" and other things?
22:25
<Philip`>
Ask Hixie :-)
22:25
<heycam>
Hixie is hereby asked
22:26
<zcorpan_>
iirc, it's to be compatible with parseInt() in as many languages as possible
22:26
<Philip`>
zcorpan_: Why would you parse a floating point number with parseInt?
22:26
<zcorpan_>
er
22:26
<zcorpan_>
parseFloat
22:27
<Philip`>
It seems much more useful to be compatible with printFloat, because millions of people are going to be doing that, whereas about five people are going to be writing web browsers that have to parse these floats
22:27
<heycam>
and they already have routines for parsing floats with "e+" in them from the js engines, i suppose
22:27
<Hixie>
please send e-mail if you want e+ support
22:29
zcorpan_
was thinking of http://www.whatwg.org/specs/web-forms/current-work/#number
22:29
<Philip`>
zcorpan_: Ah, I was thinking of http://www.whatwg.org/specs/web-apps/current-work/#rules-for-parsing-floating-point-number-values
22:29
<Philip`>
particularly when it's used for reflecting content attributes
22:36
Philip`
will send e-mail
22:40
gsnedders
often emails himself
22:40
gsnedders
is that much of a loner
22:44
Lachy
waves from London
22:44
<Lachy>
I arrived here earlier this afternoon
22:44
<gsnedders>
How come you're in London?
22:45
<Lachy>
holiday
22:45
<Lachy>
also because there is an Australia shop here where I can stock up on my supplies of Vegemite, Tim Tams, Aeroplane Jelly and Cottees Cordial
22:47
<Philip`>
Is Vegemite much different to Marmite?
22:47
<Lachy>
I just hope they have all that stuff in stock. If they don't, I'll be disappointed because the shipping costs for that stuff if I have to get it posted are expensive
22:48
<Lachy>
Vegemite is the original and superior version of Marmite
22:48
<Lachy>
I think marmite is quite disqusting, actually
22:48
<Lachy>
*disgusting
22:49
<smedero>
what are the differences in texture and/or taste?
22:49
<Lachy>
marmite is a lot more runny
22:49
<Lachy>
it's difficult to describe the differences in taste
22:50
<Lachy>
smedero, what country are you in?
22:50
<smedero>
united states
22:50
smedero
hides
22:50
<Lachy>
there are some Australia stores in the US. Order a small tube of it and try it out
22:51
<Lachy>
do you get marmite in the uS?
22:51
gsnedders
takes his pitch-fork to smedero
22:51
<Philip`>
It comes in tubes? How odd
22:51
<Lachy>
it comes in tubes and jars
22:51
<smedero>
yeah, i've come across it from time to time... more so in the northeast than here on the west coast.
22:51
<Lachy>
the tubes are smaller, lighter and cost less to post
22:52
<Lachy>
it also comes in 1.5kg tubs, which I will get if I can find it here
22:52
<gsnedders>
1.5KG!?
22:53
<Lachy>
yeah. I only saw it a few times though while I was in Aus. Though I never bought it because the smaller jars were more convenient and getting more wasn't a problem back then.
22:53
<Hixie>
larry finally replied to my e-mail on www-tag
22:54
<Hixie>
but it seems he just restated his original statement using a different pair of terms
22:54
<Hixie>
ignoring my points entirely
22:54
<gsnedders>
Why would we care about your points?
22:54
<gsnedders>
I mean, you're Hixie.
22:54
<gsnedders>
By definition, you are wrong. You're not Linus Torvalds.
22:55
<roc>
Marmite predates Vegemite
22:55
<Lachy>
woah! 2.5kg http://www.everythingaustralian.com/ve2tubainst.html
22:58
<Lachy>
roc, wow. I didn't realise that. I always though marmite was the rip off product
22:58
<roc>
of course, you're Austrlian
22:58
<roc>
you obviously think you invented the pavlova as well
22:59
<Lachy>
of course :-)
23:01
<Lachy>
wtf? Wikipedia says it may have come from New Zealand
23:01
<gsnedders>
Hah.
23:01
<Lachy>
and it also claims ANZAC biscuits came from there too. I refuse toe believe that
23:01
<gsnedders>
Australia does not rule the world.
23:01
<Lachy>
s/toe/to/
23:02
<Philip`>
I thought pavlova was designed as a treat to be used in canine psychology experiments
23:02
<Lachy>
gsnedders, sure it does
23:02
<Lachy>
well, at least we can still claim to have invented the Lamingtons
23:02
<gsnedders>
You can _claim_ whatever the hell you want.
23:02
<gsnedders>
You just probably won't be right to claim absolutely anything.
23:03
Philip`
congratulates Lachy for the invention of something he's never heard of
23:03
<Lachy>
http://en.wikipedia.org/wiki/Lamington
23:04
<Lachy>
I made a batch of those myself yesterday and brought them with me to London
23:04
<heycam>
"Tea and lamingtons are part of the festivities that follow Australian Citizenship ceremonies."
23:04
<heycam>
citation required? :)
23:05
<roc>
that seems too hokey to be true, even for Australians
23:05
<gsnedders>
That whole article seems rather devoid of citations
23:05
<Lachy>
yeah, I was surprised when I read that. But I can't confirm or deny it since I never had to go through the ceremony
23:05
<roc>
after the tea and lamingtons you have to go surfing and play "Waltzing Matilda" on a digeridoo
23:06
<Lachy>
LOL
23:07
<heycam>
one thing you do get after being conferred citizenship at a cermemony is a native australian plant
23:08
<Lachy>
heycam, is that a Gum Tree or Wattle or something else?
23:08
<heycam>
Lachy, i'm not sure, i never looked closely to see what sort of plant it was
23:08
<heycam>
(for the 2 or 3 that i've seen)
23:12
<Lachy>
heycam, those 2 plants are distinctive enough to not need a close inspection to identify
23:15
<heycam>
probably, although i don't know if a young wattle would have the bright yellow flowers to make it easier to distinguish
23:15
<Lachy>
ok
23:18
<zcorpan_>
Philip`: i'd like to look at some pages that match (maxlength|hspace|vspace|border|cols|rows|size|span|colspan|rowspan|cellpadding|cellspacing|topmargin|leftmargin|marginwidth|marginheight|scrollamount|scrolldelay|start|value|width|height)\s*=\s*["']?\s*\d+[a-zA-Z]
23:19
<Hixie>
heh
23:19
<Hixie>
i know what zcorpan is doing!
23:19
<Hixie>
and i approve
23:19
<Hixie>
please do let me know your results
23:19
<gsnedders>
Working out WTF IE is doing :P
23:20
<zcorpan_>
no, trying to figure out if we should do what mozilla does or not
23:22
<zcorpan_>
Hixie: btw, firefox does it in standards mode too... my remark about quirks was unrelated
23:23
<Hixie>
really? i fouund differences when i tested it
23:23
<Hixie>
i wonder what i was testing
23:24
<Hixie>
maybe it was safari
23:26
<zcorpan_>
it's easy to get confused while testing it (i got width/height backwards when testing moz) :)
23:32
<Philip`>
zcorpan_: Millions of people write <td width="123px"> etc - should I try excluding those?
23:33
<Philip`>
zcorpan_: Lots of people write <input value="123abc"> etc too - should I remove 'value' from the list?
23:36
<gsnedders>
Man, there is no consensus for weather forecasts for Saturday for Edi
23:37
<Philip`>
I forecast that it will be cold, and maybe wet
23:37
<gsnedders>
:P
23:37
gsnedders
is probably going to go to Edinburgh for no real reason
23:38
<gsnedders>
Just to do something different.
23:39
<roc>
zcorpan_: what are you talking about?
23:41
<Philip`>
zcorpan_: http://philip.html5.org/data/attribute-numbers-followed-by-letters.txt
23:41
<Philip`>
zcorpan_: http://philip.html5.org/data/attribute-numbers-followed-by-letters-excluding-boring-stuff.txt
23:41
<Philip`>
zcorpan_: (I can make the list longer if you wait a while)
23:41
<Hixie>
the "px" case seems like a very important one
23:42
<Hixie>
does IE really ignore all those cases?
23:42
<zcorpan_>
roc: <table width="100a"> vs <table width="100x">
23:43
<zcorpan_>
Hixie: ie doesn't do it for width/height
23:44
<Hixie>
ah
23:45
<Hixie>
then philip should exclude width/height, not "px" :-)
23:45
<Hixie>
most of the rest seem to be <font size=...pt>
23:46
<Hixie>
does it do it for size=""?
23:47
<zcorpan_>
did i write a-z? i meant a-f :/
23:47
<Philip`>
Oh
23:47
<zcorpan_>
sorry
23:47
<Philip`>
Should I try it again with [a-f]?
23:47
<Philip`>
(Should I still include 'value'?)
23:49
<zcorpan_>
if you like :) (and no)
23:50
<Philip`>
http://philip.html5.org/data/attribute-numbers-followed-by-af.txt is not many yet
23:51
Philip`
will update it in a few minutes when there's more search results
23:52
<zcorpan_>
thanks
23:55
<zcorpan_>
<font size="1em">
23:56
<zcorpan_>
height=38border=0
23:56
<Philip`>
http://philip.html5.org/data/attribute-numbers-followed-by-af.txt - now with a bit more
23:57
Philip`
leaves his grep running, since it might take a while
23:57
<zcorpan_>
<FONT SIZE="20em">
23:59
<zcorpan_>
which is http://www.angelfire.com/wi/baril/ -- looks wrong in firefox